# Bright Data MCP

Official Web MCP server from Bright Data that gives AI agents real-time, unblocked access to web search, scraping, navigation, and structured data extraction powered by the world's \#1 web data platform.

Canonical URL: https://a2a-mcp.org/entry/brightdata-mcp

Language: en

Category: MCP Servers

Updated: 2026-05-13

## Overview

The Web MCP by Bright Data is the official Model Context Protocol (MCP) server that provides an all-in-one solution for public web access. It enables LLMs, AI agents, and apps to search the web in real-time, extract data from any website, navigate and interact with pages – without getting blocked, rate-limited, or served CAPTCHAs.

Powered by Bright Data’s global proxy network and scraping infrastructure.

## Features

- Real-time web search with AI-optimized results and geo-targeting.
- Clean scraping to Markdown (including JavaScript-rendered pages).
- Browser automation and navigation (Pro mode).
- 60+ specialized tools for structured data extraction (e-commerce, social media, finance, code packages, etc.).
- AI-ranked discover search.
- Automatic anti-bot protection and proxy rotation.
- Tool groups and custom tool selection.

## Modes

- **Rapid Mode** (default, free tier): Web search, scrape_as_markdown, discover tool. 5,000 requests/month.
- **Pro Mode**: All 60+ tools including browser control and web_data_* APIs (pay-as-you-go).
- **Custom Mode**: Whitelist specific tool groups or individual tools via environment variables.

## Use Cases

- Real-time research and fact-checking agents.
- Competitor monitoring, price tracking, and lead generation.
- Autonomous browsing and form interaction.
- Structured data extraction from e-commerce, social, finance, and app stores.
- Up-to-date code/package intelligence (npm, PyPI).

## MCP Setup

**Prerequisites**: Bright Data account required. Valid API token (Admin permissions recommended for automatic zone creation). Node.js for local/self-hosted runs.

**Access Requirements**: API token required (token-based authentication via environment variable or URL query parameter). No browser OAuth or sign-in.

**Transport**: HTTP (hosted endpoint) or stdio (via npx for CLI clients).

**Docs URL**: https://brightdata.com/ai/mcp-server, https://github.com/brightdata/brightdata-mcp, https://brightdata.com/blog/ai/claude-code-web-mcp

**Hosted Option** (no install):
`https://mcp.brightdata.com/mcp?token=YOUR_API_TOKEN_HERE`

**Local Run**:
`npx @brightdata/mcp` (set `API_TOKEN` env var)

**Claude Code** (exact command from official blog):
```bash
claude mcp add brightData \
  --env API_TOKEN=<YOUR_BRIGHT_DATA_API_KEY> \
  --env PRO_MODE=true \
  -- npx -y @brightdata/mcp
```
(Windows PowerShell version uses `-- -- npx -y @brightdata/mcp`)

**Claude Desktop** (exact config block from GitHub README):
```json
{
  "mcpServers": {
    "Bright Data": {
      "command": "npx",
      "args": ["@brightdata/mcp"],
      "env": {
        "API_TOKEN": "<your-api-token-here>",
        "PRO_MODE": "true"
      }
    }
  }
}
```

**Cursor / Windsurf / Other coding agents**: Use the same npx configuration pattern or hosted HTTP URL as shown in GitHub README examples for coding agents. No unique per-client commands documented beyond the general MCP config.

**Verification**: In supported clients, run `/mcp` to confirm connection to the `brightData` server.

**Notes**: Free tier (Rapid mode) includes only base tools. Pro mode incurs usage-based charges. Environment variables control mode and limits. Source: Official GitHub README (https://github.com/brightdata/brightdata-mcp) and Claude Code integration guide (https://brightdata.com/blog/ai/claude-code-web-mcp).

## Tags

- mcp-server
- web-scraping
- browser-automation
- data-extraction
- ai-agents
- claude
- serp
- proxies

## Sources

- [Project website](https://brightdata.com/ai/mcp-server)